wattch

The description states that Wattch is:

• A trace-first measurement platform.
• Designed to give coding agents a local power meter.
• Built using MCP (Model Context Protocol) and CLI tools.
• Capable of capturing telemetry including raw samples, timestamps, source descriptors, events, command provenance, stdout, stderr, and exit status in replayable artifacts.
• Intended for use with deterministic synthetic sources, Linux RAPL, or Apple powermetrics.
• Not designed to measure provider-side model inference, private per-process energy, or whole-machine energy without required evidence.

It is built as a Rust workspace with a strict daemon/client boundary. The system includes:

• A versioned protocol carrying compact raw samples over local Unix sockets.
• A deterministic test daemon for reproducibility.
• A power daemon isolating privileged hardware access.
• An unprivileged wattch CLI handling discovery, workload capture, inspection, and validation.
• A bounded MCP server exposing the same workflow to coding agents.
• Python tools for experiment scheduling and offline analysis.

Inference The product is described as a technical tool for energy measurement in code optimization workflows. It is not presented as a commercial SaaS offering or platform with customers.

The description states:

• Wattch aims to make the energy impact of AI coding agents visible and auditable.
• It positions itself as a way to provide trustworthy local evidence for energy use in code optimization.
• The tool supports correctness-preserving loops: measure, change one thing, test, measure again, report only what the evidence supports.
• It explicitly avoids making claims about provider-side inference or whole-machine energy without supporting data.

Inference Wattch is positioned as a developer tool focused on energy transparency and measurement fidelity in AI-assisted coding. Its positioning reflects a focus on reproducibility and trustworthiness over broad applicability.

The description does not name specific customers or personas. However, it implies:

• Developers working with AI coding agents.
• Teams looking to optimize code for energy efficiency.
• Users who value correctness-preserving workflows and auditable telemetry.

Inference The target is likely developers or engineering teams using AI tools in development environments, particularly those concerned with performance, correctness, and energy impact. No evidence of segmentation or targeting beyond this general category.

The description does not contain any information about:

• Revenue streams.
• Pricing models.
• Monetization strategy.
• Customer acquisition or retention mechanisms.

Not evidenced

The description states:

• Built with Rust, Python, and technologies like clap, codex, gpt-5.6, model-context-protocol, apple-powermetrics, linux-rapl.
• Uses a daemon/client architecture with Unix sockets for communication.
• Supports deterministic synthetic sources, Linux RAPL, and Apple powermetrics.
• Includes tools for experiment scheduling and offline analysis.
• Designed to be integrated into CI pipelines.

Inference The technical stack suggests a developer-focused tool built for performance and reproducibility. It is not described as a hosted service or cloud-based solution.

The description states:

• Submitted to the OpenAI 2026 hackathon.
• Used in a controlled experiment involving two models (Luna and Terra) with 32 trials.
• All trials retained with zero gaps or missingness.
• Independent verification of artifact hashes.
• Identified a real regression in one implementation.

However, there is no evidence of:

• Revenue.
• Customers.
• Product adoption.
• Market traction.
• Commercial deployment.

Not evidenced

The description does not mention any competitors or direct market comparisons. It focuses on Wattch’s own functionality and experimental validation rather than its place in a competitive landscape.

Not evidenced

Key risks and open questions based on the self-reported description:

• The project is presented as a hackathon submission with no evidence of commercial traction or product-market fit.
• No revenue, customer data, or market validation provided.
• The tool is described as local-only and not scalable to enterprise or cloud use cases.
• There is no indication that it has been adopted by developers beyond the authors’ own experiments.

Inference The lack of any commercial evidence raises concerns about whether this is a viable product or just an experimental prototype. It also suggests limited scalability or market readiness.

1. What is the intended path from prototype to commercial product?
2. Are there any early adopters or pilot users beyond the hackathon participants?
3. How does Wattch plan to scale beyond local development environments?
4. Is there a roadmap for integrating with major AI coding platforms (e.g., GitHub Copilot, Cursor)?
5. What are the long-term plans for monetization and pricing?
6. Has the tool been tested in real-world development workflows outside of controlled experiments?